Years of Life Lost (YLL) and Disability-Adjusted Life Years (DALY) Analysis

A Comprehensive Burden of Disease Study — Kenya, 2019

Author

Timothy Achala

Published

May 2, 2026

1 Introduction and Conceptual Framework

1.1 Background

The Global Burden of Disease (GBD) framework provides a systematic approach to quantifying health loss across populations. Two foundational metrics are:

Metric	What it Measures	Components
YLL	Premature mortality burden	Deaths × remaining life expectancy
YLD	Non-fatal health burden	Prevalence × disability weight × duration
DALY	Total health burden	YLL + YLD

One DALY = one lost year of healthy life. It represents the gap between current health status and an ideal situation where everyone lives in full health to an old age.

1.2 Epidemiological Justification

1.2.1 Why YLL Matters

Years of Life Lost penalises deaths at younger ages more heavily than deaths at older ages. A child who dies at age 2 contributes far more YLL than an 80-year-old dying of the same disease. This is crucial for priority-setting because it focuses attention on preventable premature mortality.

1.2.2 Why DALY Is Superior to Simple Death Counts

Death counts ignore:

Non-fatal burden: Depression rarely kills but causes enormous suffering.
Age at death: A death at 25 is more tragic (in terms of life-years) than at 85.
Severity of disability: A 10-year episode of mild anaemia ≠ 10 years of severe HIV.

DALYs address all three limitations simultaneously.

2 Data Loading and Exploration

2.1 Load the Dataset

Code

# ─────────────────────────────────────────────────────────────────
# READ DATA
# The CSV contains 128 rows covering 6 diseases × 8 age groups × 2 sexes
# in Kenya, 2019 (GBD reference year).
# ─────────────────────────────────────────────────────────────────
raw_data <- read_csv("burden_of_disease_dataset.csv",
                     show_col_types = FALSE)

# Convert categorical variables to factors with meaningful ordering
data <- raw_data |>
  mutate(
    # Ordered age groups ensure correct axis ordering in plots
    age_group = factor(age_group, levels = c(
      "0-4", "5-14", "15-29", "30-44", "45-59", "60-69", "70-79", "80+"
    )),
    sex              = factor(sex,              levels = c("Male", "Female")),
    disease_category = factor(disease_category, levels = c(
      "Communicable", "Non-Communicable", "Cardiovascular", "Mental Health", "Injuries"
    )),
    # Flag: does this disease cause direct mortality?
    cause_of_death = as.logical(cause_of_death == "Yes"),
    # age_at_death is already numeric; NAs are read as NA automatically by read_csv
    age_at_death = suppressWarnings(as.numeric(age_at_death))
  )

# Quick confirmation
cat("Dimensions:", nrow(data), "rows ×", ncol(data), "columns\n")

Dimensions: 128 rows × 22 columns

Code

cat("Diseases:", paste(unique(data$disease_name), collapse = ", "), "\n")

Diseases: Ischemic Heart Disease, Stroke, Lower Respiratory Infections, HIV/AIDS, Diabetes Mellitus, Malaria, Depressive Disorders, Road Injuries

Code

cat("Age groups:", paste(levels(data$age_group), collapse = ", "), "\n")

Age groups: 0-4, 5-14, 15-29, 30-44, 45-59, 60-69, 70-79, 80+

Interpretation: The dataset captures 6 major disease causes (Ischemic Heart Disease, Stroke, Lower Respiratory Infections, HIV/AIDS, Diabetes Mellitus, Malaria, Depressive Disorders, Road Injuries) across 8 age strata for both sexes in Kenya 2019. The total of 128 records allows stratified analysis by age, sex, disease category, and cause type.

2.2 Data Quality Check

Code

# ─────────────────────────────────────────────────────────────────
# MISSING VALUE AUDIT
# age_at_death is legitimately NA for non-fatal diseases
# (e.g., Depressive Disorders where cause_of_death = FALSE)
# ─────────────────────────────────────────────────────────────────
missing_summary <- data |>
  summarise(across(everything(), ~sum(is.na(.)))) |>
  pivot_longer(everything(), names_to = "Variable", values_to = "Missing_n") |>
  filter(Missing_n > 0) |>
  mutate(
    Missing_pct = round(Missing_n / nrow(data) * 100, 1),
    Explanation = case_when(
      Variable == "age_at_death" ~ "Expected: non-fatal diseases have no age at death",
      TRUE ~ "Investigate further"
    )
  )

missing_summary |>
  gt() |>
  tab_header(title = "Missing Data Audit") |>
  cols_label(Variable = "Column", Missing_n = "N Missing",
             Missing_pct = "% Missing", Explanation = "Reason") |>
  tab_style(style = cell_fill(color = "#FFF3CD"),
            locations = cells_body(rows = Missing_n > 0))

Column	N Missing	% Missing	Reason
Missing Data Audit
age_at_death	2	1.6	Expected: non-fatal diseases have no age at death

Interpretation: The only missing values appear in age_at_death for Depressive Disorders, which is by design — depression is classified as primarily non-fatal in this dataset, so no age-at-death value applies. No unexpected missingness exists, confirming data integrity.

2.3 Descriptive Summary Table

Code

# ─────────────────────────────────────────────────────────────────
# SUMMARY BY DISEASE AND SEX
# Aggregates key burden indicators before YLL/DALY calculation
# ─────────────────────────────────────────────────────────────────
summary_tbl <- data |>
  group_by(disease_name, disease_category, sex) |>
  summarise(
    Total_Deaths      = sum(deaths, na.rm = TRUE),
    Total_Incident    = sum(incident_cases, na.rm = TRUE),
    Total_Prevalent   = sum(prevalent_cases, na.rm = TRUE),
    Mean_Dis_Weight   = round(mean(disability_weight), 3),
    .groups = "drop"
  ) |>
  arrange(disease_category, disease_name, sex)

summary_tbl |>
  gt() |>
  tab_header(
    title    = "Burden of Disease: Summary Statistics by Cause and Sex",
    subtitle = "Kenya, 2019"
  ) |>
  fmt_number(columns = c(Total_Deaths, Total_Incident, Total_Prevalent),
             use_seps = TRUE, decimals = 0) |>
  cols_label(
    disease_name     = "Disease",
    disease_category = "Category",
    sex              = "Sex",
    Total_Deaths     = "Total Deaths",
    Total_Incident   = "Incident Cases",
    Total_Prevalent  = "Prevalent Cases",
    Mean_Dis_Weight  = "Mean DW"
  ) |>
  tab_row_group(label = "Cardiovascular",  rows = disease_category == "Cardiovascular") |>
  tab_row_group(label = "Communicable",    rows = disease_category == "Communicable") |>
  tab_row_group(label = "Mental Health",   rows = disease_category == "Mental Health") |>
  tab_row_group(label = "Non-Communicable",rows = disease_category == "Non-Communicable") |>
  tab_row_group(label = "Injuries",        rows = disease_category == "Injuries") |>
  tab_style(style = cell_fill(color = "#E8F4FD"),
            locations = cells_row_groups())

Disease	Category	Sex	Total Deaths	Incident Cases	Prevalent Cases	Mean DW
Burden of Disease: Summary Statistics by Cause and Sex
Kenya, 2019
Injuries
Road Injuries	Injuries	Male	7,815	75,500	175,600	0.370
Road Injuries	Injuries	Female	2,320	29,750	69,450	0.370
Non-Communicable
Diabetes Mellitus	Non-Communicable	Male	13,212	39,443	116,540	0.049
Diabetes Mellitus	Non-Communicable	Female	10,730	33,157	98,287	0.049
Mental Health
Depressive Disorders	Mental Health	Male	311	39,770	118,350	0.145
Depressive Disorders	Mental Health	Female	440	59,695	178,980	0.145
Communicable
HIV/AIDS	Communicable	Male	14,610	42,400	202,600	0.547
HIV/AIDS	Communicable	Female	16,660	47,670	225,600	0.547
Lower Respiratory Infections	Communicable	Male	11,570	179,500	330,000	0.279
Lower Respiratory Infections	Communicable	Female	9,320	150,000	279,000	0.279
Malaria	Communicable	Male	8,090	382,000	678,500	0.186
Malaria	Communicable	Female	7,330	361,800	643,000	0.186
Cardiovascular
Ischemic Heart Disease	Cardiovascular	Male	15,132	30,445	83,300	0.432
Ischemic Heart Disease	Cardiovascular	Female	11,899	25,485	71,020	0.432
Stroke	Cardiovascular	Male	17,841	35,175	94,530	0.552
Stroke	Cardiovascular	Female	14,042	29,271	79,455	0.552

3 Years of Life Lost (YLL) Calculation

3.1 Theoretical Framework

YLL measures the years of life a person would have lived had they not died prematurely. The GBD uses the standard expected years of life lost formula:

\[\text{YLL}_{age,sex} = N_{deaths} \times L_{age,sex}\]

where \(L_{age,sex}\) is the remaining life expectancy at the age of death according to a standard life table (GBD uses 86.0 years for females, 91.9 years for males — here we use a unified 86-year frontier).

The remaining life expectancy for someone dying at age \(a\) is:

\[L(a) = \text{Standard Life Expectancy} - a\]

This means a 2-year-old dying contributes \((86 - 2) = 84\) YLL, while a 75-year-old dying contributes only \((86 - 75) = 11\) YLL.

3.2 YLL Computation in R

Code

# ─────────────────────────────────────────────────────────────────
# YLL CALCULATION
#
# Formula: YLL = Deaths × (Standard LE - Age at Death)
#
# Key design decisions:
# 1. We use the GBD standard LE of 86.0 years (reference frontier).
# 2. Mid-point of the age group is used as the proxy age at death
#    when age_at_death is missing (only for non-fatal diseases).
# 3. YLL is set to 0 for non-fatal diseases (cause_of_death = FALSE).
# 4. We add a floor: if (standard_LE - age_at_death) < 0, YLL = 0
#    to handle the rare case where age_at_death exceeds the standard.
# ─────────────────────────────────────────────────────────────────
STANDARD_LE <- 86.0   # GBD universal standard life expectancy (years)

yll_data <- data |>
  mutate(
    # ── Mid-point of age group for cases where age_at_death is NA ──
    age_midpoint = (age_group_lower + age_group_upper) / 2,

    # ── Use actual recorded age at death, else fall back to midpoint ──
    effective_age_at_death = coalesce(age_at_death, age_midpoint),

    # ── Remaining life expectancy at age of death ──
    remaining_LE = pmax(STANDARD_LE - effective_age_at_death, 0),

    # ── YLL = Deaths × Remaining LE ──
    # Non-fatal diseases (cause_of_death = FALSE) get YLL = 0
    YLL = if_else(cause_of_death, deaths * remaining_LE, 0),

    # ── YLL rate per 100,000 population (for comparability) ──
    YLL_rate_100k = (YLL / population) * 100000
  )

# Show a sample of the computation
yll_data |>
  filter(deaths > 0) |>
  select(disease_name, age_group, sex, deaths, effective_age_at_death,
         remaining_LE, YLL, YLL_rate_100k) |>
  slice_sample(n = 12) |>
  arrange(disease_name, age_group) |>
  kable(
    caption  = "Sample YLL Computations (12 randomly selected rows)",
    col.names = c("Disease", "Age Group", "Sex", "Deaths",
                  "Age at Death", "Remaining LE", "YLL", "YLL/100k"),
    digits   = c(0, 0, 0, 0, 1, 1, 0, 1),
    format.args = list(big.mark = ",")
  ) |>
  kable_styling(bootstrap_options = c("striped", "hover"), full_width = FALSE)

Sample YLL Computations (12 randomly selected rows)
Disease	Age Group	Sex	Deaths	Age at Death	Remaining LE	YLL	YLL/100k
Depressive Disorders	60-69	Male	38	65.2	20.8	0	0.0
Diabetes Mellitus	5-14	Male	12	9.4	76.6	919	38.3
Diabetes Mellitus	80+	Female	3,800	84.3	1.7	6,460	3,588.9
HIV/AIDS	5-14	Female	420	10.1	75.9	31,878	1,386.0
HIV/AIDS	15-29	Male	2,800	22.5	63.5	177,800	4,678.9
HIV/AIDS	80+	Male	180	83.4	2.6	468	312.0
Lower Respiratory Infections	0-4	Male	2,800	1.8	84.2	235,760	19,646.7
Lower Respiratory Infections	5-14	Male	420	9.1	76.9	32,298	1,345.8
Lower Respiratory Infections	60-69	Male	1,400	64.8	21.2	29,680	3,491.8
Lower Respiratory Infections	80+	Male	3,200	83.2	2.8	8,960	5,973.3
Stroke	0-4	Male	18	2.5	83.5	1,503	125.2
Stroke	80+	Male	6,200	83.8	2.2	13,640	9,093.3

Interpretation of YLL computation logic:

remaining_LE is the core multiplier. A child dying at age 2 carries 84 remaining years; a person dying at 75 carries only 11. This is the mathematical expression of the social preference for averting premature death.
pmax(..., 0) ensures we never obtain negative YLL, which could theoretically occur if someone dies beyond the 86-year frontier (edge case).
coalesce() gracefully handles diseases with no recorded death age by substituting the age-group midpoint — a standard epidemiological imputation.

3.3 YLL by Disease and Age Group

Code

# ─────────────────────────────────────────────────────────────────
# AGGREGATE YLL BY DISEASE AND AGE GROUP
# This is the primary YLL summary table for interpretation
# ─────────────────────────────────────────────────────────────────
yll_disease_age <- yll_data |>
  group_by(disease_name, disease_category, age_group) |>
  summarise(
    total_deaths = sum(deaths),
    total_YLL    = sum(YLL),
    YLL_rate     = sum(YLL) / sum(population) * 100000,
    .groups      = "drop"
  )

# Total YLL per disease (both sexes combined)
yll_disease_total <- yll_data |>
  group_by(disease_name, disease_category) |>
  summarise(
    Deaths   = sum(deaths),
    YLL      = sum(YLL),
    YLL_rate = sum(YLL) / sum(population) * 100000,
    .groups  = "drop"
  ) |>
  arrange(desc(YLL))

yll_disease_total |>
  gt() |>
  tab_header(
    title    = "Total YLL by Disease Cause",
    subtitle = "Kenya, 2019 — All ages, both sexes combined"
  ) |>
  fmt_number(columns = c(Deaths, YLL), use_seps = TRUE, decimals = 0) |>
  fmt_number(columns = YLL_rate, decimals = 1) |>
  cols_label(
    disease_name     = "Disease",
    disease_category = "Category",
    Deaths           = "Total Deaths",
    YLL              = "Total YLL",
    YLL_rate         = "YLL Rate (per 100k)"
  ) |>
  data_color(
    columns = YLL,
    palette = "Blues"
  ) |>
  tab_footnote(
    footnote  = "YLL Rate standardised to population size per disease stratum.",
    locations = cells_column_labels(columns = YLL_rate)
  )

Disease	Category	Total Deaths	Total YLL	YLL Rate (per 100k)¹
Total YLL by Disease Cause
Kenya, 2019 — All ages, both sexes combined
HIV/AIDS	Communicable	31,270	1,509,929	5,392.6
Malaria	Communicable	15,420	1,007,966	3,599.9
Lower Respiratory Infections	Communicable	20,890	709,141	2,532.6
Stroke	Cardiovascular	31,883	427,099	1,525.4
Road Injuries	Injuries	10,135	421,772	1,506.3
Ischemic Heart Disease	Cardiovascular	27,031	354,507	1,266.1
Diabetes Mellitus	Non-Communicable	23,942	352,611	1,259.3
Depressive Disorders	Mental Health	751	0	0.0
¹ YLL Rate standardised to population size per disease stratum.

3.4 Visualisation: YLL by Disease and Age

Code

# ─────────────────────────────────────────────────────────────────
# HEATMAP: YLL per disease × age group
# Log scale used because the range is very wide (child deaths create
# very high YLL despite fewer absolute numbers).
# ─────────────────────────────────────────────────────────────────
yll_heat <- yll_data |>
  group_by(disease_name, age_group) |>
  summarise(YLL = sum(YLL), .groups = "drop") |>
  filter(YLL > 0)

ggplot(yll_heat, aes(x = age_group, y = fct_reorder(disease_name, YLL, sum),
                     fill = YLL)) +
  geom_tile(colour = "white", linewidth = 0.5) +
  geom_text(aes(label = if_else(YLL >= 1000,
                                paste0(round(YLL / 1000, 0), "k"),
                                as.character(round(YLL)))),
            size = 3, colour = "white", fontface = "bold") +
  scale_fill_viridis_c(
    name   = "YLL",
    trans  = "log10",
    labels = label_comma(),
    option = "plasma"
  ) +
  labs(
    title    = "Years of Life Lost by Disease and Age Group",
    subtitle = "Kenya 2019 — Both sexes combined; colour on log₁₀ scale",
    x        = "Age Group", y = NULL,
    caption  = "Values > 1000 shown as 'Xk'. Source: Simulated GBD-style dataset."
  ) +
  theme_burden()

Figure 1: Heatmap of YLL by Disease and Age Group, Kenya 2019

Interpretation of heatmap:

HIV/AIDS in the 30–44 age group shows the darkest cells, reflecting the catastrophic toll of HIV on working-age adults in sub-Saharan Africa. Deaths at this age carry approximately 48 remaining life years each.
Malaria in children aged 0–4 generates enormous YLL despite lower death counts than cardiovascular diseases, purely because each death removes ~84 years of potential life.
Cardiovascular diseases (IHD, Stroke) peak in the 60–79 age bands. While death counts are highest here, YLL per death is low (6–16 years remaining), explaining their moderate YLL despite high mortality.
The near-empty cells for Depressive Disorders reflect its classification as primarily non-fatal; the few deaths are suicide-related.

3.5 YLL by Sex — Population Pyramid Style

Code

# ─────────────────────────────────────────────────────────────────
# DIVERGING BAR CHART (POPULATION PYRAMID STYLE)
# Males plotted left (negative values), females right (positive)
# This makes sex disparities immediately visible
# ─────────────────────────────────────────────────────────────────
yll_sex <- yll_data |>
  group_by(age_group, sex) |>
  summarise(YLL = sum(YLL), .groups = "drop") |>
  mutate(YLL_plot = if_else(sex == "Male", -YLL, YLL))

ggplot(yll_sex, aes(x = YLL_plot, y = age_group, fill = sex)) +
  geom_col(alpha = 0.85, width = 0.75) +
  geom_vline(xintercept = 0, colour = "black", linewidth = 0.8) +
  scale_x_continuous(
    labels = function(x) paste0(comma(abs(x / 1e3)), "k"),
    breaks = pretty(c(-max(abs(yll_sex$YLL_plot)), max(abs(yll_sex$YLL_plot))), 6)
  ) +
  scale_fill_manual(values = c("Male" = "#2166AC", "Female" = "#D6604D")) +
  labs(
    title    = "Years of Life Lost by Age Group and Sex",
    subtitle = "Males (left) vs Females (right) — Kenya 2019",
    x        = "YLL (thousands)",
    y        = "Age Group",
    fill     = "Sex",
    caption  = "Each bar represents total YLL summed across all 6 diseases."
  ) +
  theme_burden() +
  theme(legend.position = "top")

Figure 2: YLL by Age Group and Sex — Population Pyramid

Interpretation — Sex Disparities in YLL:

Males bear a heavier YLL burden in almost all age groups, particularly in the 15–44 range. This is driven by (a) higher road injury mortality in males,
1. greater cardiovascular event rates in males, and (c) higher occupational HIV exposure in some contexts.
Females aged 15–44 show relatively elevated YLL for their sex, largely due to HIV/AIDS (females account for the majority of new HIV infections in sub-Saharan Africa through heterosexual transmission).
The 60+ age groups show converging YLL between sexes, reflecting greater female longevity but higher female prevalence of stroke and diabetes at older ages.

4 Disability-Adjusted Life Years (DALY) Calculation

4.1 Conceptual Components

\[\text{DALY} = \text{YLL} + \text{YLD}\]

4.1.1 YLD (Years Lived with Disability)

\[\text{YLD} = I \times DW \times L\]

where:

\(I\) = number of incident cases in the period
\(DW\) = disability weight (0 = perfect health; 1 = death), drawn from GBD
\(L\) = average duration of the episode (years)

Alternatively: \(\text{YLD} = P \times DW\) where \(P\) = prevalent cases (used when incidence data is unavailable). We use the incidence-based approach here as recommended by GBD 2019.

4.1.2 Disability Weights Used

Disease	DW	Interpretation
IHD	0.432	Moderate-severe: chest pain, fatigue, limitation
Stroke	0.552	Severe: paralysis, speech loss, dependency
LRI	0.279	Moderate: breathlessness, fever, activity limitation
HIV/AIDS	0.547	Severe: immunosuppression, opportunistic infections
Diabetes	0.049	Mild-moderate: managed disease with complications
Malaria	0.186	Moderate: acute fever, anaemia, prostration
Depression	0.145	Mild-moderate: low mood, anhedonia, functional impairment
Road Injuries	0.370	Moderate-severe: trauma, fractures, rehabilitation

4.2 DALY Computation

Code

# ─────────────────────────────────────────────────────────────────
# STEP 1: YLD CALCULATION
#
# YLD (incidence-based) = Incident Cases × Disability Weight × Duration
#
# Why incidence-based?
# - Aligns with GBD 2019 methodology
# - Avoids double-counting prevalent cases from previous years
# - More sensitive to new episodes and intervention effects
# ─────────────────────────────────────────────────────────────────
daly_data <- yll_data |>
  mutate(
    # YLD: multiply incident cases by DW and average duration
    YLD = incident_cases * disability_weight * duration_years,

    # DALY = YLL + YLD
    DALY = YLL + YLD,

    # Per-capita rates for population-adjusted comparisons
    YLD_rate_100k  = (YLD  / population) * 100000,
    DALY_rate_100k = (DALY / population) * 100000,

    # Fraction of DALY attributable to YLL vs YLD
    YLL_fraction  = if_else(DALY > 0, YLL  / DALY, 0),
    YLD_fraction  = if_else(DALY > 0, YLD  / DALY, 0)
  )

# Verification: print a readable summary for one disease
daly_data |>
  filter(disease_name == "HIV/AIDS", sex == "Female") |>
  select(age_group, deaths, incident_cases, disability_weight, duration_years,
         YLL, YLD, DALY) |>
  kable(
    caption   = "DALY Computation Verification: HIV/AIDS — Female",
    col.names = c("Age Group", "Deaths", "Incid. Cases", "DW", "Duration (yr)",
                  "YLL", "YLD", "DALY"),
    digits    = c(0, 0, 0, 3, 2, 0, 0, 0),
    format.args = list(big.mark = ",")
  ) |>
  kable_styling(bootstrap_options = c("striped", "hover", "condensed"),
                full_width = FALSE)

DALY Computation Verification: HIV/AIDS — Female
Age Group	Deaths	Incid. Cases	DW	Duration (yr)	YLL	YLD	DALY
0-4	720	1,900	0.547	4.8	60,408	4,989	65,397
5-14	420	1,050	0.547	7.2	31,878	4,135	36,013
15-29	4,200	12,000	0.547	9.2	263,760	60,389	324,149
30-44	7,200	20,000	0.547	10.5	344,160	114,870	459,030
45-59	2,800	8,200	0.547	8.8	92,120	39,472	131,592
60-69	820	2,800	0.547	5.8	17,220	8,883	26,103
70-79	350	1,200	0.547	3.8	3,920	2,494	6,414
80+	150	520	0.547	2.5	285	711	996

Interpretation of HIV/AIDS DALY computation:

In females aged 30–44, HIV generates the highest DALYs because both YLL (high deaths × ~48 remaining years) and YLD (high incidence × DW=0.547 × ~10.5 years duration) are simultaneously maximised.
The long duration parameter for HIV (8–10+ years) substantially amplifies YLD. This is the mathematical signature of a chronic, disabling condition — even with effective ART reducing mortality, the ongoing disability burden remains substantial.

4.3 Total DALY Summary

Code

# ─────────────────────────────────────────────────────────────────
# AGGREGATE DALY TABLE — by disease, combining both sexes
# ─────────────────────────────────────────────────────────────────
daly_summary <- daly_data |>
  group_by(disease_name, disease_category) |>
  summarise(
    Deaths      = sum(deaths),
    YLL         = sum(YLL),
    YLD         = sum(YLD),
    DALY        = sum(DALY),
    DALY_rate   = sum(DALY) / sum(population) * 100000,
    YLL_pct     = round(sum(YLL) / sum(DALY) * 100, 1),
    YLD_pct     = round(sum(YLD) / sum(DALY) * 100, 1),
    .groups     = "drop"
  ) |>
  arrange(desc(DALY))

daly_summary |>
  gt() |>
  tab_header(
    title    = "DALY Burden by Disease — Kenya 2019",
    subtitle = "All ages, both sexes; sorted by total DALY burden"
  ) |>
  fmt_number(columns = c(Deaths, YLL, YLD, DALY), use_seps = TRUE, decimals = 0) |>
  fmt_number(columns = DALY_rate, decimals = 1) |>
  fmt_number(columns = c(YLL_pct, YLD_pct), decimals = 1) |>
  cols_label(
    disease_name     = "Disease",
    disease_category = "Category",
    Deaths           = "Deaths",
    YLL              = "YLL",
    YLD              = "YLD",
    DALY             = "Total DALY",
    DALY_rate        = "DALY Rate/100k",
    YLL_pct          = "YLL %",
    YLD_pct          = "YLD %"
  ) |>
  data_color(columns = DALY, palette = "Reds") |>
  tab_footnote("% of total DALY attributable to each component.",
               locations = cells_column_labels(columns = YLL_pct)) |>
  tab_style(
    style     = cell_text(weight = "bold"),
    locations = cells_body(rows = 1)
  )

Disease	Category	Deaths	YLL	YLD	Total DALY	DALY Rate/100k	YLL %¹	YLD %
DALY Burden by Disease — Kenya 2019
All ages, both sexes; sorted by total DALY burden
HIV/AIDS	Communicable	31,270	1,509,929	426,640	1,936,569	6,916.3	78.0	22.0
Malaria	Communicable	15,420	1,007,966	646	1,008,612	3,602.2	99.9	0.1
Lower Respiratory Infections	Communicable	20,890	709,141	2,392	711,533	2,541.2	99.7	0.3
Stroke	Cardiovascular	31,883	427,099	25,495	452,594	1,616.4	94.4	5.6
Road Injuries	Injuries	10,135	421,772	7,561	429,333	1,533.3	98.2	1.8
Diabetes Mellitus	Non-Communicable	23,942	352,611	32,542	385,153	1,375.5	91.6	8.4
Ischemic Heart Disease	Cardiovascular	27,031	354,507	16,794	371,301	1,326.1	95.5	4.5
Depressive Disorders	Mental Health	751	0	111,408	111,408	397.9	0.0	100.0
¹ % of total DALY attributable to each component.

5 Visualisation and Interpretation

5.1 DALY Composition — YLL vs YLD Stacked Chart

Code

# ─────────────────────────────────────────────────────────────────
# STACKED BAR CHART: Proportion of DALY from YLL vs YLD
# This is the most important diagnostic chart — it tells you
# whether a disease kills (high YLL fraction) or disables (high YLD fraction)
# ─────────────────────────────────────────────────────────────────
daly_long <- daly_summary |>
  select(disease_name, YLL, YLD, DALY) |>
  pivot_longer(cols = c(YLL, YLD), names_to = "Component", values_to = "Value") |>
  mutate(
    Proportion = Value / DALY,
    Component  = factor(Component, levels = c("YLD", "YLL"))
  )

ggplot(daly_long,
       aes(x = Value / 1e3,
           y = fct_reorder(disease_name, DALY),
           fill = Component)) +
  geom_col(alpha = 0.88, width = 0.7) +
  geom_text(
    data = daly_summary,
    aes(x = DALY / 1e3 + 20, y = disease_name,
        label = paste0(comma(round(DALY / 1e3, 0)), "k")),
    inherit.aes = FALSE, size = 3.2, hjust = 0, fontface = "bold"
  ) +
  scale_fill_manual(
    values = c("YLL" = "#D73027", "YLD" = "#4575B4"),
    labels = c("YLL" = "YLL (premature death)", "YLD" = "YLD (disability)")
  ) +
  scale_x_continuous(labels = comma, expand = expansion(mult = c(0, 0.15))) +
  labs(
    title    = "DALY Burden by Disease: Premature Death vs Disability",
    subtitle = "Kenya 2019 — Sorted by total DALY; values in thousands",
    x        = "DALYs (thousands)",
    y        = NULL,
    fill     = "DALY Component",
    caption  = "YLL = Years of Life Lost (mortality); YLD = Years Lived with Disability"
  ) +
  theme_burden() +
  theme(legend.position = "top")

Figure 3: DALY Composition: YLL vs YLD contribution by disease

Interpretation — DALY Composition:

HIV/AIDS is dominated by YLL, meaning the primary mechanism of its burden is premature death. Even with ART, untreated or late-treated HIV kills people at working ages, generating enormous life-year losses. The YLD component reflects ongoing disability in people living with HIV.
Depressive Disorders are 100% YLD (by definition in this dataset — near zero deaths), illustrating how mental health conditions can impose massive burden that is entirely invisible to mortality statistics. This is the key argument for using DALYs over death counts in health planning.
Malaria shows a mixed pattern: childhood deaths generate YLL, while recurrent non-fatal episodes in older children and adults generate YLD. In high-transmission settings, YLD from chronic anaemia and cognitive impairment can exceed YLL.
Diabetes is predominantly YLD: the disease is well-managed enough to delay mortality, but the long duration of the condition (8–11 years average in this data) with even mild disability weights accumulates substantial YLD.

5.2 DALY Rate by Age Group and Disease

Code

# ─────────────────────────────────────────────────────────────────
# FACETED LINE CHART: DALY rate across age groups, by sex
# Shows the age-pattern of burden for each disease independently
# ─────────────────────────────────────────────────────────────────
daly_age_sex <- daly_data |>
  group_by(disease_name, age_group, sex) |>
  summarise(
    DALY_rate = sum(DALY) / sum(population) * 100000,
    .groups   = "drop"
  )

ggplot(daly_age_sex, aes(x = age_group, y = DALY_rate,
                         colour = sex, group = sex)) +
  geom_line(linewidth = 1.0, alpha = 0.9) +
  geom_point(size = 2.5, alpha = 0.9) +
  facet_wrap(~disease_name, scales = "free_y", ncol = 2) +
  scale_colour_manual(values = c("Male" = "#2166AC", "Female" = "#D6604D")) +
  scale_y_continuous(labels = comma) +
  labs(
    title    = "Age-Specific DALY Rate by Disease and Sex",
    subtitle = "Kenya 2019 — Rate per 100,000 population (y-axis free scale)",
    x        = "Age Group", y = "DALY Rate (per 100,000)",
    colour   = "Sex",
    caption  = "Note: y-axes differ across panels to show within-disease age patterns."
  ) +
  theme_burden() +
  theme(
    axis.text.x   = element_text(angle = 45, hjust = 1, size = 8),
    legend.position = "top"
  )

Interpretation — Age-Specific DALY Patterns:

Malaria shows a characteristic U-shape or monotonically declining pattern — the highest DALY rates in children under 5, then falling as acquired immunity develops, with a slight uptick in the elderly due to immune senescence. This underscores why under-5 malaria prevention (bed nets, chemoprevention) is the highest-impact intervention.
HIV/AIDS peaks dramatically in the 15–44 age band, with females consistently higher than males in the 15–29 group. This sex reversal is a hallmark of the sub-Saharan African epidemic and reflects higher female biological susceptibility and social vulnerability to HIV.
Cardiovascular diseases (IHD, Stroke) show the expected J-shaped increase with age, confirming that cardiovascular risk accumulates with ageing. Male rates exceed female rates until the 70–80+ band, where female longevity results in more years of exposure.
Road Injuries exhibit a distinctive peak in the 15–44 male age group — the classic pattern of young male risk-taking driving transport fatalities. Female rates are substantially lower across all ages.
Depression shows peak DALY rates in the 15–44 age group, especially in females aged 15–29, aligning with established epidemiology of major depressive disorder peaking in early adulthood.

5.3 DALY Treemap by Category

Code

# ─────────────────────────────────────────────────────────────────
# PROPORTIONAL STACKED BAR: Category-level DALY composition
# Useful for national health system budget allocation discussions
# ─────────────────────────────────────────────────────────────────
category_daly <- daly_data |>
  group_by(disease_category) |>
  summarise(
    DALY     = sum(DALY),
    YLL      = sum(YLL),
    YLD      = sum(YLD),
    .groups  = "drop"
  ) |>
  mutate(
    pct      = DALY / sum(DALY) * 100,
    label    = glue("{disease_category}\n{round(pct,1)}%\n({comma(round(DALY/1e3))}k)")
  ) |>
  arrange(desc(DALY))

ggplot(category_daly,
       aes(x = reorder(disease_category, DALY), y = DALY / 1e3,
           fill = disease_category)) +
  geom_col(width = 0.65, alpha = 0.88) +
  geom_text(aes(label = paste0(round(pct, 1), "%")),
            hjust = -0.2, fontface = "bold", size = 4) +
  scale_fill_brewer(palette = "Set2") +
  scale_y_continuous(labels = comma, expand = expansion(mult = c(0, 0.18))) +
  coord_flip() +
  labs(
    title    = "Total DALY Burden by Disease Category",
    subtitle = "Kenya 2019 — Percentage of total DALY shown",
    x        = NULL, y = "DALYs (thousands)",
    caption  = "Categories: GBD disease classification groupings."
  ) +
  theme_burden() +
  theme(legend.position = "none")

Figure 5: Total DALY by Disease Category — Proportional Bar Chart

5.4 YLL vs YLD Scatter — Disease Positioning

Code

# ─────────────────────────────────────────────────────────────────
# SCATTER PLOT: YLL (x) vs YLD (y)
# Diseases in the upper-left are high disability, low mortality
# Diseases in the lower-right are high mortality, low disability
# Diseases in the upper-right impose dual burden
# ─────────────────────────────────────────────────────────────────
ggplot(daly_summary,
       aes(x = YLL / 1e3, y = YLD / 1e3,
           colour = disease_category, size = DALY / 1e3)) +
  geom_point(alpha = 0.80) +
  geom_text_repel(aes(label = disease_name), size = 3.5, fontface = "bold",
                  max.overlaps = 15, box.padding = 0.5) +
  geom_abline(intercept = 0, slope = 1, linetype = "dashed",
              colour = "grey50", linewidth = 0.8) +
  annotate("text", x = max(daly_summary$YLL / 1e3) * 0.1,
           y = max(daly_summary$YLD / 1e3) * 0.9,
           label = "YLD > YLL\n(Disability-dominant)", colour = "grey40",
           size = 3.2, hjust = 0) +
  annotate("text", x = max(daly_summary$YLL / 1e3) * 0.75,
           y = max(daly_summary$YLD / 1e3) * 0.08,
           label = "YLL > YLD\n(Mortality-dominant)", colour = "grey40",
           size = 3.2, hjust = 0) +
  scale_size_continuous(name = "Total DALY (k)", range = c(3, 12)) +
  scale_colour_brewer(name = "Disease Category", palette = "Dark2") +
  scale_x_continuous(labels = comma) +
  scale_y_continuous(labels = comma) +
  labs(
    title    = "Disease Characterisation: Mortality Burden vs Disability Burden",
    subtitle = "Position relative to 45° line indicates dominant burden component",
    x        = "YLL (thousands) — Premature mortality component",
    y        = "YLD (thousands) — Disability component",
    caption  = "Dashed line = equal YLL and YLD. Point size = total DALY burden."
  ) +
  theme_burden()

Figure 6: YLL vs YLD Scatter Plot: Disease Characterisation

Interpretation — Disease Positioning:

Diseases above the 45° line (YLD > YLL) are primarily disabling rather than fatal. In this dataset, Depressive Disorders falls decisively above the line — it has near-zero YLL but substantial YLD.
Diseases below the 45° line (YLL > YLD) are primarily lethal. HIV/AIDS and Road Injuries in working-age adults fall in this zone — the main pathway of their burden is premature death, not chronic disability.
Malaria sits near the line, reflecting its dual nature: a major killer of children (YLL) and a chronic debilitating illness in older groups (YLD).
This positioning diagram is a powerful policy communication tool: diseases in the upper-left require disability management services; diseases in the lower-right require mortality prevention interventions.

6 Advanced Analyses

6.1 Population-Attributable Fraction by Age (YLL Contribution)

Code

# ─────────────────────────────────────────────────────────────────
# WHICH AGE GROUPS CONTRIBUTE MOST YLL?
# This guides intervention targeting:
# If children dominate → focus on paediatric prevention
# If working ages dominate → focus on adult screening and treatment
# ─────────────────────────────────────────────────────────────────
yll_age_paf <- yll_data |>
  group_by(age_group) |>
  summarise(
    YLL       = sum(YLL),
    Deaths    = sum(deaths),
    .groups   = "drop"
  ) |>
  mutate(
    YLL_pct   = YLL / sum(YLL) * 100,
    Deaths_pct = Deaths / sum(Deaths) * 100,
    # Ratio: how much of YLL does this age contribute relative to its share of deaths?
    # Ratio > 1 means YLL is disproportionately high (young deaths)
    YLL_Death_ratio = YLL_pct / Deaths_pct
  )

yll_age_paf |>
  kable(
    caption   = "YLL and Death Attribution by Age Group — All diseases",
    col.names = c("Age Group", "YLL", "Deaths", "YLL %", "Deaths %",
                  "YLL/Death Ratio"),
    digits    = c(0, 0, 0, 1, 1, 2),
    format.args = list(big.mark = ",")
  ) |>
  kable_styling(bootstrap_options = c("striped", "hover"), full_width = FALSE) |>
  row_spec(which(yll_age_paf$YLL_Death_ratio > 1.5), background = "#FDECEA")

YLL and Death Attribution by Age Group — All diseases
Age Group	YLL	Deaths	YLL %	Deaths %	YLL/Death Ratio
0-4	1,244,810	14,770	26.0	9.2	2.84
5-14	314,223	4,103	6.6	2.5	2.58
15-29	718,230	11,466	15.0	7.1	2.11
30-44	941,126	19,715	19.7	12.2	1.61
45-59	655,825	19,925	13.7	12.4	1.11
60-69	481,253	22,720	10.1	14.1	0.71
70-79	347,622	30,923	7.3	19.2	0.38
80+	79,937	37,700	1.7	23.4	0.07

Interpretation — YLL/Death Ratio:

A YLL/Death ratio > 1 means that age group is “over-represented” in YLL relative to its share of deaths. Highlighted rows (ratio > 1.5) indicate age groups where each death removes many remaining life years.
Children under 5 have the highest ratio: each death removes ~84 years, so even modest numbers of childhood deaths dominate total YLL.
Ages 70–80+ have a ratio < 1: they account for the majority of deaths but each death removes only 6–16 years, so their share of YLL is proportionally much smaller.
This is the mathematical justification for prioritising child health interventions in YLL-based burden analyses.

6.2 Sex-Stratified DALY Rate Comparison

Code

# ─────────────────────────────────────────────────────────────────
# DUMBBELL PLOT: Male vs Female DALY rates per disease
# Shows the magnitude and direction of sex inequality in burden
# ─────────────────────────────────────────────────────────────────
sex_daly <- daly_data |>
  group_by(disease_name, sex) |>
  summarise(
    DALY_rate = sum(DALY) / sum(population) * 100000,
    .groups   = "drop"
  ) |>
  pivot_wider(names_from = sex, values_from = DALY_rate) |>
  mutate(
    gap       = Male - Female,
    dominant  = if_else(gap >= 0, "Male higher", "Female higher")
  )

ggplot(sex_daly) +
  geom_segment(aes(x = Female, xend = Male,
                   y = fct_reorder(disease_name, Male),
                   yend = fct_reorder(disease_name, Male),
                   colour = dominant),
               linewidth = 1.8, alpha = 0.7) +
  geom_point(aes(x = Female, y = disease_name), colour = "#D6604D", size = 4) +
  geom_point(aes(x = Male,   y = disease_name), colour = "#2166AC", size = 4) +
  geom_text(aes(x = Male, y = disease_name,
                label = paste0("Δ=", round(abs(gap), 0))),
            nudge_y = 0.28, size = 3, fontface = "italic") +
  scale_colour_manual(
    values = c("Male higher" = "#2166AC", "Female higher" = "#D6604D")
  ) +
  scale_x_continuous(labels = comma) +
  labs(
    title    = "DALY Rate Disparity Between Males and Females",
    subtitle = "Dumbbell plot — Blue = Male; Red = Female; Δ = absolute difference",
    x        = "DALY Rate per 100,000 Population",
    y        = NULL,
    colour   = "Which sex has higher burden",
    caption  = "Rates standardised within each sex's own population."
  ) +
  theme_burden() +
  theme(legend.position = "top")

Figure 7: Sex-Specific DALY Rate Comparison by Disease

Interpretation — Sex Inequalities:

Road Injuries show the largest male excess, consistent with global data showing 3:1 male-to-female road mortality ratios in low- and middle-income countries. Interventions targeting young male drivers (helmet laws, breathalysing, speed enforcement) would close this gap most effectively.
HIV/AIDS in this dataset shows female excess in DALY rates due to the disproportionate HIV burden on young women in sub-Saharan Africa. This is a policy-critical finding — female-focused PrEP programmes, social protection, and economic empowerment are the highest-impact responses.
Depression shows female excess, consistent with the well-established 2:1 female-to-male ratio in major depressive disorder globally.
Cardiovascular diseases show male excess, particularly IHD, reflecting earlier age of onset in males and higher hypertension prevalence.

6.3 Cause-Deleted Life Expectancy (Hypothetical Analysis)

Code

# ─────────────────────────────────────────────────────────────────
# CAUSE-DELETED LIFE EXPECTANCY
#
# Question: How much would life expectancy increase if we
# eliminated each disease?
#
# Approximation: ΔLE ≈ YLL / total_population
# This gives the average life-years gained per person if the
# disease were fully eliminated. It is an approximation of the
# Arriaga decomposition method.
# ─────────────────────────────────────────────────────────────────
total_pop <- daly_data |>
  distinct(age_group, sex, disease_name, population) |>
  # Use one disease to avoid duplicate population rows
  filter(disease_name == "Ischemic Heart Disease") |>
  summarise(total_pop = sum(population)) |>
  pull(total_pop)

cause_deleted <- yll_data |>
  group_by(disease_name) |>
  summarise(total_YLL = sum(YLL), .groups = "drop") |>
  mutate(
    delta_LE_years = total_YLL / total_pop,
    delta_LE_days  = delta_LE_years * 365.25,
    interpretation = case_when(
      delta_LE_years >= 1    ~ "Major impact (≥1 year gain)",
      delta_LE_years >= 0.25 ~ "Moderate impact (3–12 months gain)",
      TRUE                   ~ "Modest impact (<3 months gain)"
    )
  ) |>
  arrange(desc(delta_LE_years))

cause_deleted |>
  gt() |>
  tab_header(
    title    = "Hypothetical Gain in Life Expectancy if Disease Eliminated",
    subtitle = glue("Estimated based on total YLL and population of {comma(total_pop)}")
  ) |>
  fmt_number(columns = total_YLL,      use_seps = TRUE, decimals = 0) |>
  fmt_number(columns = delta_LE_years, decimals = 3) |>
  fmt_number(columns = delta_LE_days,  decimals = 1) |>
  cols_label(
    disease_name    = "Disease",
    total_YLL       = "Total YLL",
    delta_LE_years  = "ΔLE (years)",
    delta_LE_days   = "ΔLE (days)",
    interpretation  = "Impact Level"
  ) |>
  data_color(columns = delta_LE_years, palette = "YlOrRd") |>
  tab_footnote(
    footnote  = "ΔLE ≈ YLL / population. Assumes independence of causes (not strictly valid for correlated risks).",
    locations = cells_column_labels(columns = delta_LE_years)
  )

Disease	Total YLL	ΔLE (years)¹	ΔLE (days)	Impact Level
Hypothetical Gain in Life Expectancy if Disease Eliminated
Estimated based on total YLL and population of 28,000,000
HIV/AIDS	1,509,929	0.054	19.7	Modest impact (<3 months gain)
Malaria	1,007,966	0.036	13.1	Modest impact (<3 months gain)
Lower Respiratory Infections	709,141	0.025	9.3	Modest impact (<3 months gain)
Stroke	427,099	0.015	5.6	Modest impact (<3 months gain)
Road Injuries	421,772	0.015	5.5	Modest impact (<3 months gain)
Ischemic Heart Disease	354,507	0.013	4.6	Modest impact (<3 months gain)
Diabetes Mellitus	352,611	0.013	4.6	Modest impact (<3 months gain)
Depressive Disorders	0	0.000	0.0	Modest impact (<3 months gain)
¹ ΔLE ≈ YLL / population. Assumes independence of causes (not strictly valid for correlated risks).

Interpretation — Cause-Deleted Analysis:

Eliminating HIV/AIDS would yield the largest life expectancy gain, reflecting its concentration of deaths in the 15–44 age group where remaining life expectancy is highest.
Cardiovascular diseases (IHD + Stroke combined) would provide the second largest gain. Even though cardiovascular deaths are concentrated in older ages (reducing per-death YLL), the sheer volume of deaths creates a substantial aggregate impact.
Malaria elimination would provide meaningful but smaller gains in life expectancy because, while child deaths are devastating in YLL terms, the population base is large.
Depression contributes zero ΔLE in this model because it is classified as non-fatal — reinforcing that using LE alone would entirely miss the burden of mental illness.

7 Policy Implications and Priority Setting

7.1 Cost-Effectiveness Framing

Code

# ─────────────────────────────────────────────────────────────────
# PRIORITY MATRIX
# Combines DALY burden with intervention availability data
# (Notional cost-effectiveness values for illustration)
# ─────────────────────────────────────────────────────────────────
priority_df <- daly_summary |>
  select(disease_name, disease_category, DALY, DALY_rate) |>
  mutate(
    # Notional cost per DALY averted (USD) — illustrative values based on DCP3
    cost_per_DALY_averted = c(185, 220, 1850, 6200, 4500, 95, 280, 750),
    # Intervention availability (1=good, 0=poor)
    intervention_strength = c(3, 2, 4, 2, 3, 5, 2, 3),
    priority_score = (DALY / max(DALY)) * (1000 / cost_per_DALY_averted) *
                     (intervention_strength / 5),
    priority_tier = cut(priority_score,
                        breaks   = quantile(priority_score, c(0, 0.33, 0.67, 1)),
                        labels   = c("Low Priority", "Medium Priority", "High Priority"),
                        include.lowest = TRUE)
  )

ggplot(priority_df,
       aes(x = DALY / 1e3,
           y = cost_per_DALY_averted,
           size = DALY_rate,
           colour = priority_tier)) +
  geom_point(alpha = 0.80) +
  geom_text_repel(aes(label = disease_name), size = 3.2,
                  box.padding = 0.5, max.overlaps = 15) +
  scale_y_log10(labels = dollar) +
  scale_size_continuous(name = "DALY Rate/100k", range = c(3, 14)) +
  scale_colour_manual(
    name   = "Priority Tier",
    values = c("High Priority"   = "#D73027",
               "Medium Priority" = "#F46D43",
               "Low Priority"    = "#74ADD1")
  ) +
  labs(
    title    = "Disease Priority Matrix: DALY Burden vs Cost-Effectiveness",
    subtitle = "High burden + low cost per DALY averted = highest priority (illustrative data)",
    x        = "Total DALY (thousands)",
    y        = "Cost per DALY Averted (USD, log scale)",
    caption  = "Cost-per-DALY values are illustrative, based on DCP3 order-of-magnitude estimates."
  ) +
  theme_burden()

Interpretation — Priority Matrix:

Diseases in the lower-right quadrant (high DALY burden, low cost per DALY averted) represent the strongest case for immediate investment. Malaria and LRI typically occupy this space because bed nets, vaccines, and antibiotics are highly cost-effective.
HIV/AIDS has high DALY but moderate cost-effectiveness — ART is effective but expensive at scale, placing it in the medium-priority tier in this simplified model. Prevention (PrEP, condom promotion) would shift it toward higher priority.
Cardiovascular and diabetes interventions are often more expensive per DALY averted due to the need for lifelong medication and monitoring, placing them in the upper portions of the plot.
This matrix is a starting point for investment cases, not a final answer. Equity considerations, political feasibility, and co-benefits must also inform resource allocation.

8 Summary Statistics and Final Table

Code

# ─────────────────────────────────────────────────────────────────
# COMPREHENSIVE FINAL SUMMARY TABLE
# Suitable for a national health report or policy brief appendix
# ─────────────────────────────────────────────────────────────────
final_table <- daly_data |>
  group_by(disease_name, disease_category, sex) |>
  summarise(
    Deaths     = sum(deaths),
    YLL        = sum(YLL),
    YLD        = sum(YLD),
    DALY       = sum(DALY),
    DALY_rate  = sum(DALY) / sum(population) * 100000,
    YLL_frac   = round(sum(YLL) / sum(DALY) * 100, 1),
    .groups    = "drop"
  ) |>
  arrange(disease_category, disease_name, sex)

# Render as interactive DataTable for exploration
datatable(
  final_table |>
    mutate(across(c(Deaths, YLL, YLD, DALY), ~comma(round(.))),
           DALY_rate = round(DALY_rate, 1),
           YLL_frac  = paste0(YLL_frac, "%")),
  colnames  = c("Disease", "Category", "Sex", "Deaths", "YLL", "YLD",
                "DALY", "DALY Rate/100k", "YLL%"),
  filter    = "top",
  extensions = "Buttons",
  options   = list(
    pageLength = 16,
    dom        = "Bfrtip",
    buttons    = c("copy", "csv", "excel"),
    scrollX    = TRUE
  ),
  caption   = "Full DALY Results Table — Kenya 2019 (Interactive: search, sort, export)"
)

9 Methodological Notes

9.1 Key Analytical Assumptions

Analytical Assumptions and Justifications
Assumption	Value_Used	Justification
Standard Life Expectancy	86.0 years (GBD 2019 standard frontier)	Represents the ideal life expectancy achievable — aspiration rather than current LE
YLD Formula	Incidence × Disability Weight × Duration	Preferred over prevalence-based when incidence data available
Age at Death Proxy	Mid-point of age group when not recorded	Standard epidemiological convention for grouped data
Population Denominator	Population size per age-sex stratum	Ensures rates are comparable across strata of different sizes
Discounting	No discounting applied (consistent with GBD 2010+)	GBD dropped discounting in 2010 to avoid systematically undervaluing elderly burden
Age-weighting	No age-weighting (GBD 2010+ methodology)	GBD dropped age-weighting in 2010 to avoid systematically undervaluing children/elderly

9.2 Limitations

Simulated data: This dataset is designed for methodological illustration. Real GBD analyses use cause-of-death models, DisMod-MR for disease modelling, and systematic reviews — not point estimates.
Independence assumption: The cause-deleted LE analysis assumes diseases are independent, which is not true (e.g., diabetes increases cardiovascular and infectious disease risk simultaneously).
Comorbidity not modelled: Individuals with multiple conditions experience disability from each, but joint disability weights are not additive — the GBD uses a multiplicative correction.
No uncertainty intervals: Real GBD estimates include 95% uncertainty intervals derived from Monte Carlo simulation of input uncertainty.

10 References and Further Reading

GBD 2019 Diseases and Injuries Collaborators (2020). Global burden of 369 diseases and injuries in 204 countries and territories, 1990–2019. The Lancet, 396(10258), 1204–1222.
Murray CJL, Lopez AD (1996). The Global Burden of Disease. Harvard School of Public Health / WHO.
Salomon JA et al. (2012). Disability weights for the Global Burden of Disease 2010 study. The Lancet, 380(9859), 2129–2143.
Institute for Health Metrics and Evaluation (IHME). GBD Compare tool. http://vizhub.healthdata.org/gbd-compare
Disease Control Priorities 3rd Edition (DCP3). http://dcp-3.org

Code

sessionInfo()
## R version 4.3.1 (2023-06-16 ucrt)
## Platform: x86_64-w64-mingw32/x64 (64-bit)
## Running under: Windows 11 x64 (build 26100)
## 
## Matrix products: default
## 
## 
## locale:
## [1] LC_COLLATE=English_Kenya.utf8  LC_CTYPE=English_Kenya.utf8   
## [3] LC_MONETARY=English_Kenya.utf8 LC_NUMERIC=C                  
## [5] LC_TIME=English_Kenya.utf8    
## 
## time zone: Africa/Nairobi
## tzcode source: internal
## 
## attached base packages:
## [1] stats     graphics  grDevices utils     datasets  methods   base     
## 
## other attached packages:
##  [1] DT_0.34.0          glue_1.8.0         RColorBrewer_1.1-3 viridis_0.6.5     
##  [5] viridisLite_0.4.2  gt_1.3.0           janitor_2.2.1      patchwork_1.3.2   
##  [9] ggrepel_0.9.8      scales_1.4.0       kableExtra_1.4.0   knitr_1.50        
## [13] lubridate_1.9.4    forcats_1.0.1      stringr_1.5.2      dplyr_1.2.0       
## [17] purrr_1.1.0        readr_2.1.5        tidyr_1.3.1        tibble_3.3.0      
## [21] ggplot2_4.0.0      tidyverse_2.0.0   
## 
## loaded via a namespace (and not attached):
##  [1] gtable_0.3.6      bslib_0.9.0       xfun_0.53         htmlwidgets_1.6.4
##  [5] tzdb_0.5.0        vctrs_0.7.2       tools_4.3.1       crosstalk_1.2.2  
##  [9] generics_0.1.4    parallel_4.3.1    pkgconfig_2.0.3   S7_0.2.0         
## [13] lifecycle_1.0.5   compiler_4.3.1    farver_2.1.2      textshaping_1.0.5
## [17] snakecase_0.11.1  htmltools_0.5.8.1 sass_0.4.10       yaml_2.3.10      
## [21] jquerylib_0.1.4   pillar_1.11.1     crayon_1.5.3      cachem_1.1.0     
## [25] tidyselect_1.2.1  digest_0.6.37     stringi_1.8.7     labeling_0.4.3   
## [29] fastmap_1.2.0     grid_4.3.1        cli_3.6.5         magrittr_2.0.4   
## [33] withr_3.0.2       bit64_4.6.0-1     timechange_0.3.0  rmarkdown_2.30   
## [37] bit_4.6.0         gridExtra_2.3     hms_1.1.4         evaluate_1.0.5   
## [41] rlang_1.1.7       Rcpp_1.1.1        xml2_1.4.1        svglite_2.2.2    
## [45] rstudioapi_0.18.0 vroom_1.6.6       jsonlite_2.0.0    R6_2.6.1         
## [49] systemfonts_1.3.2 fs_1.6.6